The report reviews the literature concerning written driver license examinations which shows that current written examinations are poor predictors of unsafe drivers. Although some studies demonstrate significant relationships between written examination scores and accidents these relationships are significant only for drivers with certain combinations of sex, age and level of education. The conclusions are based on tests which are currently used and which have been criticized as not clearly testing knowledge or not being statistically reliable. A reason to introduce written examinations can be to counteract information deficiencies which can be effective if drivers can be classified into groups with identified information deficiencies. The report also reviews classifications with identified information deficiencies and suggests how further research can be conducted with properly developed examinations.
Abstract